Human Capital and Economic Opportunity: A Global Working Group. Working Paper Series. Working Paper No.

Transcription

1 Human Capital and Economic Opportunity: A Global Working Group Working Paper Series Working Paper No. Human Capital and Economic Opportunity Working Group Economic Research Center University of Chicago 1126 E. 59th Street Chicago IL humcap@uchicago.edu

2 Personality Structure 1 Running Head: PERSONALITY STRUCTURE THE STRUCTURE OF TEMPERAMENT AND PERSONALITY TRAITS: A DEVELOPMENTAL PERSPECTIVE Rebecca L. Shiner Colgate University Colin G. DeYoung University of Minnesota To appear in the Oxford Handbook of Developmental Psychology

3 Personality Structure 2 THE STRUCTURE OF TEMPERAMENT AND PERSONALITY TRAITS: A DEVELOPMENTAL PERSPECTIVE Introduction Humans show a panoply of individual differences in their typical behavior, emotions, and thoughts. Beginning in infancy, individuals vary in traits such as energy and activity level, positive emotional engagement with others, feelings of distress and irritability, and persistent attention and interest in absorbing tasks. Older children, adolescents, and adults vary in their typical self-discipline, responsibility, empathy, imagination, and intellect. Traits show some stability across time and situations, but they also change over time and show some degree of situational specificity (McAdams & Pals, 2006). Contemporary research on temperament and personality traits addresses fundamental questions about these individual differences: What are the biological and environmental sources of variation in traits? To what extent and how do traits remain the same and change over time? How do individuals traits affect their physical and mental health, relationships, work, and well-being? These questions are best answered when researchers can achieve some consensus about the basic structure of traits. A structure or taxonomy of traits articulates which traits covary with which other traits, which traits are the most important, and what form those traits take at various points in the life course. A trait structure thus provides an organizational scheme for the basic units of temperament and personality and identifies how those basic units relate to one another. In the past, students of personality development used a bewildering array of measures and scales to describe individual differences, with the unfortunate consequence that results were difficult to compare from one study to the next. A structure or taxonomy provides a number of benefits in the study of personality development. First, researchers can make greater progress when they use similar language to study the same traits. As with psychiatric diagnoses, communication is aided by a common language for describing the phenomena of interest. Second, a taxonomy of traits enables integration of new findings with previous research. New measures can be related to the known structure of traits. A shared taxonomy can prevent the re-christening of alreadyrecognized individual differences. New findings for each trait can be integrated with existing nomological networks for those traits, and in turn new hypotheses can be generated. Although taxonomies provide structure for research, they can likewise be modified by new findings; taxonomies are organizational systems that evolve over time.

4 Personality Structure 3 Because children s traits expand so much in number and variety over the early years, it is challenging to develop a clear taxonomy or structure for describing these traits and their patterns of co-occurrence. As children grow from infancy to the school-age years, they develop new capacities, and these capacities greatly increase the number of traits that children can display. Within the first year of life, children already display temperamental differences in positive emotions and pleasure, various negative emotions, and interest and attention (Rothbart & Bates, 2006). As children move out of infancy, the development of more coordinated motor skills enables children to display physical aggression (Tremblay & Nagin, 2005) and to explore more widely. Children change from manifesting only a small number of emotions during infancy pleasure, distress, and interest to manifesting an expanded set of emotions by age 3 including joy, sadness, anger, fear, empathy, pride, shame, and guilt (Eisenberg, 2000; Lewis, 2000). Children s capacities for self-regulation likewise develop rapidly, which enables children to display differences in their abilities to regulate their emotions, engage in moral behavior, and pursue tasks (Eisenberg, Hofer, & Vaughan, 2007). In short, children s maturation permits the development and expression of new personality traits. The more narrow range of temperament traits seen in infants expands into a more complex network of traits in the school-age years. During each phase of early development, the structure is likely to be different, as new traits become apparent. Despite these challenges, over the last two decades, substantial progress has been made in identifying the structure of temperament and personality traits during each phase of life from early childhood through adulthood. One important finding emerging from recent work on temperament and personality structure is that these individual differences are organized hierarchically across the lifespan. Some specific behavioral descriptors tend to co-vary (e.g., talkative, expressive, not shy when meeting new people). The co-variation among those descriptors is explained by lower-order traits that are relatively narrow in focus (e.g., sociability or assertiveness). In turn, some lowerorder traits tend to co-vary, and the co-variation among those traits is accounted for by higherorder traits with greater breadth (e.g., Extraversion). Traits manifest a hierarchical structure like this in infancy and early childhood (Rothbart & Bates, 2006), middle childhood and adolescence (Caspi & Shiner, 2006), and adulthood (DeYoung, Quilty, & Peterson, 2007; Digman, 1997; Markon, Krueger, & Watson, 2005). As we discuss in more detail later in the chapter, there is newer evidence that, at an even higher level, these higher-order traits tend to show reliable

5 Personality Structure 4 patterns of co-variation, forming metatraits (DeYoung, 2006; Digman, 1997). Thus, traits demonstrate a hierarchical structure that ranges from lower-order traits to higher-order traits to metatraits. In this chapter, we articulate a developmental perspective on personality traits from early childhood through adulthood. In the first section, we address two topics that are fundamental in defining the most important traits at each point in the life span: the relationship between temperament and personality and the methods used to ascertain the structure of traits in the temperament and personality research traditions. We argue in this section that temperament and personality are different ways of describing the same basic traits, with temperament research primarily focused on early-emerging individual differences and personality research focused on individual differences that appear later in childhood and continue into adulthood. In the second section, we describe the current status of the most prominent models of temperament, as well as the most widely-accepted personality trait model, the Big Five (John, Naumann, & Soto, 2008). In the third section, we articulate a structural model that integrates contemporary findings on temperament and personality traits from early childhood through adulthood. We use the Big Five trait structure, along with one additional childhood trait, to organize this taxonomy. In the fourth section, we discuss the current research on the psychological and biological processes that underlie individual differences in the Big Five traits in childhood and adulthood. In the final sections, we offer concluding thoughts on the nature of personality trait development and suggestions for future research. Definitions and Measurement: Temperament and Personality Traits Although people display individual differences in traits across the lifespan, these traits are described sometimes as temperament and other times as personality. Before turning to a discussion of the trait structures found in childhood and adulthood, it is important to address two issues that impact the structures obtained for temperament and personality traits: the definitions of temperament and personality and the means used to measure traits and ascertain the structure of traits. Researchers definitions of temperament and personality affect which traits or descriptors they choose to include in various models of individual differences, and the traits included in different models in turn affect the resulting structure. What are temperament and personality?

6 Personality Structure 5 The concept of temperament has a long history, beginning in ancient times. Notions of temperament date at least from the ancient Greek idea that a person's typical mood and behavior result from the balance of four humors in the body: blood, black bile, yellow bile, and phlegm. In this ancient model, temperament was viewed as deriving from biological and emotional processes, a view consistent with current conceptualizations of temperament (Clark & Watson, 2008; Zuckerman, 1995). In more recent times, the empirical study of temperament in childhood was galvanized by the work of Alexander Thomas and Stella Chess, who started a longitudinal study of children s early-emerging behavioral styles in 1956 (Thomas, Chess, Birch, Hertzig, & Korn, 1963). At the time that Thomas and Chess began their study, most research on personality development was based on the assumption that children s socialization experiences were the most important sources of their individual differences in personality. Thomas and Chess work helped to convince researchers, practitioners, and parents that children vary biologically from one another from early in life and that these biological differences are important for the course of children s development. In the years since Thomas and Chess sparked interest in childhood temperament, the amount of research on the topic has grown at a rapid pace. Different models of temperament have been put forth, and these structural models will be discussed in the following section. At present, temperament researchers and practitioners are not yet unanimous in their definition of temperament. Perhaps the most influential contemporary definition of temperament comes from Mary Rothbart. Rothbart and colleagues argue that temperament includes individual differences in affect, activity, attention, and self-regulation (Rothbart & Bates, 2006). Like the ancient Greek model, this model highlights the importance of individual differences in emotional processes, including many different positive and negative emotions; these differences reflect children s reactivity to the environment. Unlike the Greek model, this model equally emphasizes the importance of individual differences in the regulation of reactive tendencies through attention and other aspects of self-regulation. According to this contemporary view, temperamental traits emerge during childhood, are closely linked with biological processes, and are in part shaped by heredity; however, experience also shapes their development. This model captures most of the points of agreement among current temperament models (Zentner & Bates, 2008). Personality includes a much broader range of individual differences than does temperament. McAdams and Pals (2006) have developed a particularly helpful model for

7 Personality Structure 6 understanding the purview of personality, which they divide into three broad levels: traits, characteristic adaptations, and personal narratives. First, traits describe relatively stable patterns of behavior, motivation, emotion, and cognition (Pytlik Zillig, Hemenover, & Dienstbier, 2002; Wilt & Revelle, 2009) that are not bound to a particular sociocultural context but could be observed in any such context. This is not to say that all traits will be evident to the same extent in all cultures, nor that all traits can be observed in any situation, but rather that any trait can be observed in a subset of situations in any culture, regardless of time and place. Second, characteristic adaptations include a wide range of motivational, social-cognitive, and developmental adaptations that are specific to a particular time, place, or role (p. 208). For example, youths vary in their goals and their sense of competence and self-efficacy in particular domains of their lives (e.g., academics, friendships) (Shiner, in press). Third, by adolescence youths begin to form personal narratives that help them to make sense of their identities and selves over time (McAdams, 2008). These narratives are unique to each person but can be studied empirically in terms of their common features across individuals. This chapter addresses traits, while acknowledging that personality extends well beyond that level, even in childhood (Shiner, in press). Historically, temperament and personality have been studied as distinct sets of individual differences, with temperament consisting of more narrowly defined consistencies that appear earlier in life and with personality consisting of a broader range of consistencies that emerge later in life. However, if we restrict our consideration of personality to traits rather than characteristic adaptations or narratives, then temperament and personality traits have much in common (see Caspi & Shiner, 2006, Clark & Watson, 2008, McCrae et al., 2000, and Zentner & Bates, 2008 for similar arguments that personality traits in adulthood are, in essence, temperamental traits). First, both sets of individual differences are shaped by heredity and by the environment (Krueger & Johnson, 2008; Saudino, 2005). A common assumption about traits is that they start at birth as largely heritable in origin and gradually come to be more influenced by the environment, as children have more and more life experiences. However, this assumption is incorrect, in part because it overlooks the fact that, before a child s birth, the intrauterine environment has already influenced the expression of each child s genetic material (Feldman, 2008). In addition, there is evidence that early traits sometimes become more rather than less related to genetics as children grow from infancy to childhood (Knafo & Plomin, in press;

8 Personality Structure 7 Saudino, 2005). Temperament traits in childhood and personality traits in adulthood both follow another interesting pattern: Stability in individuals traits seems to derive from genetic influences, whereas changes in traits are influenced by both genetic and environmental factors (Ganiban, Saudino, Ulbricht, Neiderhiser, & Reiss, 2008; Krueger & Johnson, 2008; Saudino, 2005). In short, current behavior genetic research makes clear that temperament and personality traits both arise from the complex interplay of genes and experiences. Second, animals display individual differences in behavior that mirror most of the major temperament dimensions in childhood and personality trait dimensions in adults (Weinstein, Capitanio, & Gosling, 2008). Temperament and personality traits thus may be more elaborated forms of basic behavioral systems that appear across species. These temperament and personality traits may reflect individual differences in biological systems that have been selected through evolution and are shaped by individuals life experiences (Nettle, 2006). There are a number of biological systems that are relevant for personality functioning and that that are crucial for human survival for example, systems supporting the detection of rewards and threats, achievement of social dominance, striving after long-term goals, nurturance of the young, aggression, and exploration of new environments. Although such biological systems are part of the human make-up, people vary in the strength and expression of such systems. Individuals life experiences create further variations in the expression of these systems, which become manifest in traits across the life span. Third, both temperament traits and personality traits show both stability and change over time (Roberts & Del Vecchio, 2000). Sometimes child psychologists and psychiatrists shy away from using the term personality to describe individual differences in children, because of understandable but incorrect assumptions about the nature of personality traits. Although personality is often assumed to be extremely stable across time and situations and essentially unchangeable, more recent work suggests that personality differences, in transaction with environmental circumstances, organize behavior in dynamic ways over time (Caspi & Shiner, 2006). Fourth, a final point of convergence between temperament and personality is their very similar structure and content, as reviewed in detail later in this chapter. Research on the structure of temperament and personality traits proceeded in two distinct traditions; despite this, the two lines of research have converged on two similar sets of traits. Thus, it may be helpful to view temperament and personality not as truly distinct forms of individual differences, but rather as

9 Personality Structure 8 different ways of describing the same basic traits, with temperament typically referring to earlier forms of these traits and personality to later forms. Because of the substantial overlap between temperament and personality traits, the two sets of individual differences are discussed together throughout this review. How is structure established for temperament and personality traits? Given that temperament and personality traits have often been conceptualized as different constructs, it is not surprising that these two domains of individual differences have been measured differently. As noted, the way that traits are measured has important implications for the way structure is established. Temperament research is marked by great variety in the ways that temperament traits are measured. This richness of methods is likely due to several underlying factors. The first reason is straightforward: Children cannot easily report on their own traits, because of the limitations of their language development and self-insight. Child psychologists have not had the luxury of simply asking their subjects to describe themselves and have had to employ more complex methods. Second, children s behavioral tendencies can probably be observed more easily and more naturally than adults. Infants and very young children are less aware of being observed than adults are, and even preschoolers are likely to be less self-conscious and more spontaneous than adults while being observed in natural contexts or during lab tasks. Third, child psychologists have recognized for many years that children s behavior varies across context and that different informants have different insights into children s behaviors (Achenbach, McConaughy, & Howell, 1987). Consequently, temperament researchers have relied on multiple methods and reporters to assess children s traits (Caspi & Shiner, 2008; Rothbart & Bates, 2006). As with adults, questionnaires are often used; parents are the most frequent reporters, but teachers, older children, and adolescents can also complete questionnaires. Naturalistic observations are used to code children s behavioral tendencies. For example, home observation systems have been developed to assess individual differences in preschoolers (e.g., Buckley, Klein, Durbin, Hayden, & Moerk, 2002). Laboratory tasks create specific situations in which children s behaviors can be observed. Tasks have been developed to assess specific individual differences, such as Effortful Control (Kochanska & Knaack, 2003) and behavioral inhibition (Kagan, Snidman, & Arcus, 1998). A more comprehensive battery of laboratory tasks assessing temperament in children is

10 Personality Structure 9 available and widely used (Goldsmith, Reilly, Lemery, Longley, & Prescott, 1995). These varied methods have allowed temperament researchers to explore children s individual differences in a more valid fashion than would be possible with reliance on a single method. Although these varied methods have been a strength of temperament research, some challenges have made it difficult to establish a clear structure of temperamental differences in childhood. With some of the methods described, the researcher has to begin with a reasoned hypothesis about the crucial temperament traits. For example, in conducting naturalistic observations or creating lab tasks, the researcher must start with some notion of the relevant traits in order to create the coding scheme or to construct the lab task. Empirical means can be used to refine the measurement of those traits, but the basic boundaries of the traits must be defined at the outset; this limits the usefulness of these methods as a means of establishing trait structure. Naturalistic observations and behavioral tasks also share the problem of potentially tapping more than one underlying trait, which makes it difficult to interpret the meaning of any particular task. For example, behavioral observations of inhibition in young children may confound temperamental differences in low levels of positive emotions in response to novelty and high levels of negative emotions in response to novelty (Putnam & Stifter, 2005); children may be inhibited because of either one or both of these tendencies. Questionnaires may provide an easier means of establishing trait structure, because they can identify which traits tend to co-occur across a much wider range of traits more quickly. However, temperament questionnaires have been constructed with an eye toward measuring specific temperament models. Thus, as with naturalistic observations and lab tasks, researchers begin with reasoned hypotheses about the relevant traits and construct the questionnaires to measure those traits. The researchers guiding assumptions about the basic nature of temperament also limit the traits sampled. Thus, any particular temperament questionnaire is likely to fail to include all potential traits. Because of the limitations of the methods used for assessment, there remains no general consensus about the structure of temperament in childhood. However, as we hope to demonstrate in our discussion of temperament structure, despite the varied means of measuring temperament, some traits have emerged as important across models and measures. The fundamental difficulty in investigating trait structure is to create a sufficiently broad and unbiased pool of trait measurements, in which to identify structure. A reasonably

11 Personality Structure 10 representative sample from the universe of all possible traits must be used to ensure unbiased results in factor analysis, which identifies broad patterns of covariation among variables. The lexical approach, which samples trait-descriptive words from natural language, is the best approximation of this criterion yet developed. Another valuable approximation is to sample traits from a large number of existing questionnaires (e.g., Markon et al., 2005). Neither approach ensures a complete lack of bias in the pool of traits, but they are less biased than other methods that have been used for trait selection. Research on personality structure in childhood and adulthood provides a striking contrast to temperament research. In adulthood, self-report questionnaires have reigned as the most common method of personality assessment by far. Unlike younger children, adults are assumed to possess insights into their typical thoughts, behaviors, and feelings. Self-report questionnaires are used in part because they are inexpensive and easy to administer, but they have more substantive strengths as well. Questionnaires aggregate information about behavior across a number of situations and over a period of time; they efficiently gather a lot of information about a wide variety of traits; and they can solicit information about relatively rare but important behaviors. Although work on trait structure in adults has tended to rely on individuals selfreports, the reports of others have been used as well, including spouses and peers (Costa & McCrae, 1992; McCrae, Jang, Livesley, Riemann, & Angleitner, 2001; Riemann, Angleitner, & Strelau, 1997). Many adult personality measures include a large number of person descriptors that sample a very wide range of behaviors. By employing this kind of questionnaire study, in conjunction with factor analysis, over the last several decades, personality researchers have made large strides in identifying the basic structure of personality traits in adults (John et al., 2008). This structure, known as the Five Factor Model or the Big Five, will be described in detail later in the chapter; the Big Five traits include Extraversion, Neuroticism, Conscientiousness, Agreeableness, and Openness/Intellect. The research on personality structure in childhood has been far less extensive than that in adults. As we will describe in the section on personality structure in childhood, personality questionnaires describing children and adolescents yield evidence of a Big Five trait structure for youth, even as early as preschool-age (Caspi & Shiner, 2006). The methods used to measure temperament and personality and to establish trait structure have offered different strengths to these research traditions. Because personality psychologists

12 Personality Structure 11 have devoted considerable effort to establishing trait structure through extensive empirical analysis of multiple questionnaires, the field has converged on a generally accepted trait structure. As a consequence of this general agreement about structure, adult personality research has made great strides in exploring the nature of traits for example, their genetic and biological bases, cross-cultural manifestations, and underlying psychological processes. Because the study of temperament has focused instead on the development of different temperament models, there is less consensus among temperament researchers about the general structure of temperament. The field of temperament has made tremendous progress in recent years. But, even greater progress is likely to occur as temperament researchers empirically test and evaluate competing models of individual differences and arrive at greater consensus about the basic units of temperament. Still, the multiple methods used to study temperament have been a great strength of this research tradition. Temperament research puts in bold relief the overreliance of personality research on questionnaires, particularly self-report questionnaires. There have been a number of personality studies examining personality through behavioral observations; one of the best examples is the recent German Observational Study of Adult Twins (Borkenau, Riemann, Angleitner, & Spinath, 2001), which has provided important insights into the genetic and environmental sources of variation in adults observed traits. Nonetheless, such studies remain the exception. Kagan (2003) has been a particularly vocal critic of overreliance on self-report questionnaires, stating, Conclusions about a child s psychological features based only on questionnaires or interviews have a meaning that is as limited as Ptolemy s conclusions about the cosmos based on the reports of observers staring at the night sky without a telescope (p. Kagan, 2003). This claim is too strong, given the substantial knowledge accumulated through the use of questionnaires and interviews in both children and adults. Nonetheless, as Kagan (2007) notes, work on individual differences will proceed faster and better as researchers turn to more varied methods. In short, the temperament and personality traditions have much to offer each other in the study of personality development. Models of Temperament and Personality Structure In this section, we describe several temperament models and the most widely accepted model of personality. Because temperament researchers have not reached consensus on what model provides the most valid structure, we review the most prominent models of temperament

13 Personality Structure 12 that have been developed over the last several decades. As noted in the previous section, personality researchers have generally converged on the Big Five model for describing trait structure in adults (John et al., 2008), and there is increasing evidence that this model captures the structure of personality traits in children as well (Caspi & Shiner, 2006). Thus, we present the evidence for this personality model only, rather than reviewing alternative personality models. This brief review of various temperament and personality structural models highlights points of convergence in the traits included in various models. Thomas and Chess As noted, Alexander Thomas and Stella Chess galvanized contemporary interest in temperament among both researchers and clinicians. Thomas and Chess (1977) guided the wellknown New York Longitudinal Study (NYLS) of temperament, based on their overarching framework for temperament: We conceptualize temperament as the stylistic component of behavior that is the how of behavior as differentiated from motivation, the why of behavior, and abilities, the what of behavior (Goldsmith et al., p. 508). In other words, they wanted to focus on behavioral style the variations in how children display their behavior. They presumed that such differences would have, in part, an endogenous biological basis, given their emergence early in infancy. Thomas and Chess derived their list of temperament traits based on a content analysis of a small set of interviews with parents in the NYLS study. The list of traits included nine dimensions (Thomas & Chess, 1977; Thomas et al., 1963): (a) activity level; (b) regularity of biological functions, (c) initial approach or withdrawal from new stimuli, (d) adaptability to new situations following the initial response, (e) threshold of sensory responsiveness, (f) intensity of emotional reactions (regardless of quality), (g) general positivity versus negativity of mood, (h) distractibility, or capacity for external stimuli to alter behavior, and (i) attention span or persistence in the face of obstacles. These traits were chosen with an eye toward identifying traits with likely impact on later functioning. Later work has identified some conceptual and empirical problems with the Thomas and Chess model. Conceptually, it is not truly possible to distinguish children s style of behavior from the content of and motivation for their behavior; what children do cannot be disentangled from how they do it, and motivation influences both the content and style of behavior. Empirically, the Thomas and Chess-inspired questionnaires do not yield nine distinct temperament traits (De Pauw, Mervielde, & Van Leeuwen, 2009; Shiner & Caspi, 2003), but

14 Personality Structure 13 rather a much smaller number of traits. These traits include social inhibition, irritability, attention/persistence, activity level, and sensory sensitivity (Martin, Wisenbaker, & Huttenen, 1994). Despite the problems that have become evident with newer research, the Thomas and Chess model successfully inspired contemporary research on temperament and highlighted aspects of young children s behavior that have long-term clinical implications. Buss and Plomin In contrast to Thomas and Chess s focus on traits appearing in infancy, Arnold Buss and Robert Plomin chose to focus their temperament model on childhood traits that were likely to be apparent from infancy through adulthood (Buss & Plomin, 1975, 1984; Goldsmith et al., 1987). They argued that temperament traits should be substantially heritable, apparent in primates, and relatively stable over time, even in childhood. When Buss and Plomin first presented their model, they believed that four traits fit these criteria, and these form the acronym EASI: emotionality (focused on negative emotions, first undifferentiated distress and later both fear and anger), activity, sociability, and impulsivity. Sociability was later differentiated from shyness, with the former tapping a preference for interaction with others and the latter tapping discomfort interacting with unfamiliar people. Impulsivity was moved out of the model and later added back in as empirical work explored the construct. The EASI model has fallen out of favor in research on childhood temperament, probably because it leaves out some traits that could reasonably be considered temperamental in nature and thus paints too narrow a picture of temperament. Nonetheless, this model pointed researchers to the importance of understanding traits that appear both early and later in life and identified some of the most important traits that appear across models. Rothbart Mary Rothbart s theoretical model of temperament was described earlier in this chapter, because it is the model that most often guides current research on temperament. As noted, Rothbart has argued that temperament traits consist of constitutional differences in reactivity and self-regulation, with constitutional seen as the relatively enduring biological makeup of the organism influenced over time by heredity, maturation, and experience (Rothbart & Derryberry, 1981, p. 37). Children s reactive traits (such as emotional tendencies) reflect biological arousability, whereas regulatory traits modulate children s reactivity. According to this view, new temperament traits emerge over time as children mature.

15 Personality Structure 14 In order to capture the expansion and development of temperament traits during each phase of life, Rothbart and colleagues developed questionnaire measures to assess temperament in infancy, early childhood, preschool-age, middle childhood, early adolescence, and adulthood (Putnam, Ellis, & Rothbart, 2001). In order to develop the lower-order scales at each age, Rothbart and colleagues considered other temperament models (including the NYLS and EASI models), adult temperament and personality models, and research on basic emotions (Izard, 1977). These lower-order scales were designed to assess a wide range of differences in reactivity and self-regulation at each age and have been refined empirically, although their basic content was defined rationally. Factor analyses of Rothbart and colleagues questionnaire measures yield evidence for three overarching traits from infancy through later childhood (Rothbart & Bates, 2006). Surgency taps tendencies toward high activity, a rapid approach style, expressions of positive emotions, and pleasure and excitement in social interaction. Negative Emotionality taps children s tendencies toward sadness, fear, irritability and frustration, and difficulty with being quieted after high arousal. Effortful Control (named Orienting/regulation in infancy) includes the ability to sustain attention and inhibit behavior, the ability to persist in tasks, pleasure in low intensity situations, and sensitivity to perceptual experiences. However, when Rothbart s Children s Behavior Questionnaire was factor analyzed together with a broader collection of traits in a large sample of preschool children, low intensity pleasure and perceptual sensitivity formed a separate factor, distinct from Effortful Control (De Pauw et al., 2009). Table 1 presents items assessing the three widely recognized higher-order traits in children ages 3 to 7, as well as items reflecting this additional factor. Rothbart has obtained evidence for a fourth trait, Affiliativeness, in adolescence (Rothbart & Bates, 2006) and has developed a temperament model for adults that includes a fifth trait, Orienting Sensitivity, which corresponds to the factor found by De Pauw and colleagues in children (Evans & Rothbart, 2007). The five adult temperament factors correspond closely to the Big Five (Evans & Rothbart, 2007). Rothbart s work has highlighted important higher-order traits that show clear conceptual links with personality traits observed in children and adults. Kagan Jerome Kagan s work on temperament stands in contrast to the broader temperament models outlined thus far in that he has tended to argue for a narrower definition of temperament. Specifically, Kagan (2008) has suggested that a temperamental bias refers to a biologically

16 Personality Structure 15 based foundation for clusters of feelings and subsequent actions that appear during early childhood... (p. 39) and that it is assumed, but not yet proven, that the biological foundations for many, but probably not all, human temperaments are heritable neurochemical profiles (p. 39). Rather than attempting to describe the possible full range of temperament traits, Kagan has adopted a more inductive approach by focusing in-depth on a particular observable tendency specifically, a predisposition toward high or low reactivity to novel or unfamiliar situations (Kagan, 2008; Kagan & Fox, 2006). This tendency, sometimes termed inhibition to the unfamiliar, indexes variations in the tendency to withdraw and express fear in the face of stressful novel situations (Fox, Henderson, Marshall, Nichols, & Ghera, 2005). Kagan and colleagues have obtained evidence for both the biological underpinnings and long-term outcomes of this temperamental trait (Kagan, Snidman, Kahn, & Towsley, 2007). Cloninger The models reviewed thus far have originated from attempts to understand temperament in childhood. Like Rothbart, others have considered the manifestations of temperament in adults. The most widely known theory is that of Cloninger (1987; Cloninger, Svrakic, & Przybeck, 1993) who developed a model of personality that distinguishes between four temperament traits (Novelty Seeking, Harm Avoidance, Reward Dependence, and Persistence) and three character traits (Self-Directedness, Cooperativeness, and Self-Transcendence). He hypothesized that the temperament traits would be evident early in ontogeny and strongly genetically determined. In contrast, he hypothesized that the character traits would develop later because they are determined by experience during development rather than primarily by genes. However, empirical work has revealed flaws in Cloninger s model. First, his distinction between temperament and character appears untenable. The character traits show similar levels of heritability to the temperament traits (Ando et al., 2004; Gillespie, Cloninger, Heath, & Martin, 2003). Second, Cloninger s seven-factor structure has not proven consistently replicable (Ando et al., 2004; Ball, Tennen, & Kranzler, 1999; Herbst, Zonderman, McCrae, & Costa, 2000). In fact, Cloninger s Temperament and Character Inventory is best described by the five-factor structure of the Big Five (Markon et al., 2005; Ramanaiah, Rielage, & Cheng, 2002). Harm Avoidance and Self-Determination (reversed) are both markers of Neuroticism. Cooperativeness, Persistence, and Self-Transcendence are markers of Agreeableness, Conscientiousness, and Openness/Intellect, respectively. Reward Dependence combines Agreeableness and

17 Personality Structure 16 Extraversion. And Novelty Seeking is most strongly associated with Conscientiousness (reversed), but also consistently loads positively on Extraversion as well as sometimes negatively on Agreeableness and positively on Openness/Intellect. Cloninger s model has been popular, in part, because he proposed hypotheses for the biological substrates of three of the temperament traits. As biological theories are developed to explain better-validated trait models (e,g,, DeYoung & J. R. Gray, 2009), reliance on a poorly validated model becomes less appealing. The Big Five Model from Childhood Through Adulthood One of the great achievements in the study of adult personality over the past two decades is greater clarity about the higher-order structure of personality. As noted earlier in the chapter, the most widespread support has been obtained for a five-factor structure, dubbed the Big Five or the five-factor model (Caspi & Shiner, 2006; John et al., 2008) and including broad traits of Extraversion, Neuroticism, Conscientiousness, Agreeableness, and Openness-to- Experience/Intellect. Table 1 presents items measuring these five traits in both children and adults. Support for this model derives from two main sources of evidence. First, research in the lexical tradition has examined the structure of natural language. According to the lexical hypothesis guiding this research, the personality terms contained in the natural language may provide an extensive, yet finite, set of attributes that people who share that language have found to be important and useful in their interactions with each other. Factor analyses of adjectives drawn from dictionaries in numerous countries have resulted in factors resembling the Big Five traits; strongest support has been found in languages from northern European origins (Caspi & Shiner, 2006). Recently, a six-factor solution has been discovered in lexical research that appears to be more widely replicable across languages than the Big Five (Ashton et al., 2004); however, this model appears to be only a minor variation on the Big Five, splitting Agreeableness into two factors (DeYoung et al., 2007; McCrae & Costa, 2008). Natural languages are likely to include many words describing traits within the Agreeableness domain because Agreeableness describes the quality of social interactions (social salience is one likely biasing factor in variable selection in the lexical approach). Second, factor analyses of questionnaires designed to measure a broad range of individual differences yield the Big Five traits, even when those questionnaires were not designed to assess the Big Five (Caspi & Shiner, 2006; John et al., 2008). A number of studies have jointly analyzed several personality questionnaires at once, rather than focusing on a single

18 Personality Structure 17 personality model, and these studies have provided converging evidence of this same basic structure for adult personality (e.g., Markon et al., 2005). There is now convincing evidence that, at least by the school-age years, children s personality traits share the same Big Five structure as adult traits (Caspi & Shiner, 2006; Mervielde, De Clercq, De Fruyt, & Van Leeuwen, 2005; Shiner & Masten, 2008). This fivefactor structure of children s traits has been found in studies with both parents and teachers as reporters and in both questionnaire and Q-sort measures (see Caspi & Shiner, 2006 for relevant studies). Although some of the studies obtaining a five-factor structure in childhood have employed measures pre-structured to reflect the Big Five traits, other studies have found similar structures in measures designed simply to tap a broad range of personality traits in childhood (e.g., John, Caspi, Moffitt, Robins, & Stouthamer-Loeber, 1994; Digman & Shmelyov, 1996; Shiner, 2000; Tackett et al., 2008). In fact, one of the most seminal early papers documenting a five-factor personality trait structure included teacher reports on schoolchildren s traits, using a broad, unselected set of descriptors (Digman & Takemoto-Chock, 1981). Parents reports on their children s traits show structural continuity of the Big Five traits by the time children are school-age (DeFruyt et al., 2006). Remarkably, when children as young as 6-years-old rate their personalities in the context of a interview with puppets, they can provide coherent, differentiated reports on the Big Five traits (Measelle, John, Ablow, Cowan, & Cowan, 2005). A recent study of youths ages 10 to 20 demonstrated that youths personality self-reports increasingly conform to a Big Five factor structure with age (Soto, John, Gosling, & Potter, 2008). Taken together, these studies suggest that the Big Five model can be used as an overarching taxonomy for both children s and adults personality traits. An Integrative Model of Temperament and Personality Trait Structure In this section, we describe a structural model that integrates existing research on temperament and personality traits from infancy through adulthood. Although work on temperament and personality structure has proceeded using varied models and methods, there is considerable convergence on a core set of traits across the lifespan. We organize this structure using the Big Five trait model, with the addition of one trait, activity level, that does not clearly map onto the Big Five in childhood. We chose to use the Big Five as an organizing structure, even for childhood traits, for several reasons. First, single childhood personality traits have been studied for decades

19 Personality Structure 18 including dominance, shyness, stress reactivity, aggression, delay of gratification, empathy, and achievement striving (Shiner, 1998). The Big Five model encompasses most of these single traits and relates them to each other in a structural system. Second, the Big Five model provides excellent coverage of traits that parents from many countries consider important in describing their children. As part of an international project, parents from the United States, China, Poland, Belgium, Holland, Germany, and Greece were asked to describe their 2- to 13-year old children (Kohnstamm, Halverson, Mervielde, & Havill, 1998). The vast majority of the phrases parents used to characterize their children could be easily classified as fitting into one of the Big Five trait domains. Third, the Big Five traits appear to encompass the temperament traits that emerge from multiple methods, including questionnaires, observations, and lab tasks. Factor analyses of parent and teacher questionnaires point consistently to a set of core temperament traits; these questionnaires derive from the models described in the previous section and from other models as well (e.g., Goldsmith, 1996; Kochanska, Coy, Tjebkes, & Husarek, 1998; Lemery, Goldsmith, Klinnert, Mrazek, 1999). Particularly helpful are the rare studies that examine questionnaires from multiple temperament models within the same study, because they provide evidence about the joint structure of temperament traits across models (e.g., De Pauw et al., 2009; Lemery et al., 1999). Evidence for the structure of temperament traits also derives from home observational coding systems (Bornstein, Gaughran, & Homel, 1986; Buckley, Klein, Durbin, Hayden, & Moerk, 2002) and from laboratory tasks (Goldsmith et al., 1995; Kochanska & Knaack, 2003; Majdandžić & van den Boom, 2007), which demonstrate which behavioral tendencies cohere as temperament traits. Questionnaire, observational, and lab task studies all yield a set of temperament traits that show conceptual and empirical relationships with many of the Big Five traits (see Caspi & Shiner, 2006, Mervielde & Asendorpf, 2000, and Zentner & Bates, 2008 for similar lists of traits). In these cases, the Big Five traits encompass the essential aspects of the temperament traits and add additional information. In what follows, we describe in some detail each of the traits as they are manifested from early childhood through adulthood. We begin with one narrower childhood trait activity level and then turn to the Big Five traits. In our discussion of the Big Five, we explain what personality research adds to the temperament conception of traits, and we review briefly longitudinal research linking children s early traits with the later manifestation of those traits. To

20 Personality Structure 19 illustrate the meaning of the Big Five traits, Table 1 lists items from three measures: the Children s Behavior Questionnaire (Rothbart, Ahadi, Hershey, & Fisher, 2001 ), a parent-report temperament questionnaire for children ages 3 to 7; a teacher-report instrument used to explore the structure of personality in a sample of children ages 8 to 10 (Digman & Shmelyov, 1996); the Big Five Aspect Scales (DeYoung et al., 2007), an adult Big Five questionnaire created from the International Personality Item Pool, which is in the public domain (Goldberg, 1999). After describing the various traits, we end this section with a review of metatraits, the higher-order factors that capture the covariance of the Big Five traits. Activity Level Activity level is an important component of most childhood temperament models and is typically conceptualized as the vigor and tempo of children s motor movements. Activity level emerges as a separate temperament trait in questionnaire studies from infancy through later in childhood (Caspi & Shiner, 2006; DePauw et al., 2009; Lemery et al., 1999), but it also can be measured reliably across natural and laboratory settings through the use of an actigraph, a mechanical device tracking children s movements (Saudino & Zapfe, 2008). The meaning of children s activity level is likely to change with development. Motor movement in infancy is associated with both anger and positive emotions, whereas motor movement in the toddler years is linked in complex ways with early markers of high Extraversion and low self-control (Eaton, 1994). Most childhood temperament questionnaire measures of activity level skew toward the measurement of poorly regulated, impulsive activity (De Pauw et al., 2009), which is most likely to be associated with low Conscientiousness and low Agreeableness, given that these two traits tap aspects of self-regulation. In contrast, high levels of energy, enthusiasm, and positive engagement are clearly a component of Extraversion in toddlers, older children, and adults (Goldberg, 2001; John et al., 2008; Lamb, Chuang, Wessels, Broberg, & Hwang, 2002), and Activity, in this sense, has been conceived explicitly as a facet of Extraversion (Costa & McCrae, 1992). Positive Emotionality/Extraversion Children display variations in their positive emotions in infancy; as children age, this trait broadens in content to become the trait of Extraversion. From infancy, children vary in their expression of positive emotions, including smiling and laughter, pleasure, and joy and excitement in social interactions (Gartstein & Rothbart, 2003). Both observational and